CN114268124A

CN114268124A - Distributed power supply credible capacity evaluation method based on equal power supply reliability

Info

Publication number: CN114268124A
Application number: CN202111413832.2A
Authority: CN
Inventors: 孙冰; 陈家浩; 李云飞; 曾沅; 张文旭
Original assignee: Tianjin University
Current assignee: Tianjin University
Priority date: 2021-11-25
Filing date: 2021-11-25
Publication date: 2022-04-01
Anticipated expiration: 2041-11-25
Also published as: CN114268124B

Abstract

The invention discloses a distributed power supply credible capacity evaluation method based on equal power supply reliability, which comprises the following steps: s1, determining the installed capacity of the distributed power supply; s2, establishing a model of a distributed power source, a power grid element and a load in the system; s3, generating a system time sequence running state; s4, analyzing fault consequences based on intelligent island division; s5, calculating a reliability index based on a sequential Monte Carlo method; and S6, carrying out credible capacity search based on the chord cutting method. The method is used for evaluating the credible capacity of the distributed power supply based on the equal power supply reliability principle, and aiming at the characteristic that a power distribution network can flexibly change a topological structure through a tie switch, an island division method based on a greedy algorithm is introduced into reliability calculation, so that the power supply reliability of a system is accurately evaluated through a scientific island division scheme, and the evaluation result of the credible capacity is more accurate.

Description

Distributed power supply credible capacity evaluation method based on equal power supply reliability

Technical Field

The invention belongs to the technical field of capacity planning of distributed power supplies of active power distribution networks, and relates to a credible capacity evaluation method of a distributed power supply of a power distribution network, in particular to a credible capacity evaluation method of a distributed power supply of a power distribution network based on equal power supply reliability indexes.

Background

In order to cope with environmental problems such as air pollution, global warming, etc., clean energy sources such as wind energy, light energy, etc. have been vigorously developed. With the wide access of distributed power supplies such as wind power, photovoltaic and the like to a power distribution network, the power supply capacity of a power grid is improved. However, the output of the distributed power supply such as wind and light is uncertain, and the capacity value of the distributed power supply cannot be fully considered in power system planning. In order to effectively measure the power supply capacity of the wind and light power supply, the academic world provides 'credible capacity' as an evaluation index. From a power reliability perspective, the trusted capacity may be defined as: on the premise of reliability, the system can increase the supplied load after the wind-solar power supply is connected.

Experts and scholars at home and abroad make a great deal of research on the credible capacity assessment, and a plurality of effective calculation methods are provided. The distributed power supply credible capacity evaluation mainly comprises 3 links of wind-solar power supply output modeling, system reliability calculation and credible capacity search. Different methods and models have been adopted in various links in the prior art. From the angle of a wind-solar power supply output model, the wind-solar power supply output model can be divided into a multi-state unit model, a probability density model, a time sequence output simulation model and the like; from the perspective of reliability calculation, the method can be classified into a convolution method, a Monte Carlo simulation method, a general generative function method and the like; from the perspective of a search method of the credible capacity, the search method can be divided into methods such as a dichotomy method, a chord cutting method, a simplified Newton method, an intelligent algorithm and the like. Existing researches are often used for carrying out credible capacity evaluation on a large wind-solar electric field, distributed control and multipoint access of a distributed power supply are difficult points, and the credible capacity evaluation on the distributed power supply is rarely carried out.

The core of the credible capacity evaluation of the distributed power supply lies in the reliability evaluation of the power system, and the difficulty of the evaluation is how the distributed power supply continues to supply power to the load during the system fault, so that the islanding is the basis for calculating the reliability of the power system. The essence of the islanding problem is that the distributed power supply with limited capacity is fully utilized to obtain the optimal power islanding with the distributed power supply as the center, and in the islanding mode, a part of users of the power distribution network are only powered by the distributed power supply to ensure the uninterrupted power supply of important loads in the islanding, so that the power supply reliability of the system is improved, a reasonable islanding scheme is formulated, the power supply reliability of the system can be obviously improved, and the potential of the distributed power supply is better exerted.

The existing island division solving method is abundant and can be roughly divided into the following types: 1) the minimum spanning tree algorithm, with Prim and Kruskal algorithms being the most common; 2) a heuristic algorithm, which is to determine the starting sequence and the recovery path of the power supply in the power failure region by adopting a genetic algorithm, a particle swarm algorithm and the like, and calculate to obtain the optimal solution of the island division scheme under the corresponding objective function; 3) the knapsack problem algorithm converts an island division method into a common knapsack or tree knapsack problem solution. No matter which algorithm is used for solving the islanding problem, most of the existing methods model the power distribution network into a tree model, neglect the existence of a tie switch in the power distribution network, and only perform islanding according to the topological structure of the power distribution network before the fault occurs. This may reduce the number of power supply paths selected to simplify the calculation, but also may lose the better islanding scheme.

Through searching, no prior art document which is the same as or similar to the prior art document is found.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a distributed power supply credibility assessment method based on equal power supply reliability, which is characterized in that the core is that a forward-looking greedy algorithm-based island division method is adopted to calculate the reliability of a system, and a string cutting method is adopted to search the credibility of the distributed power supply on the premise of equal reliability; the invention comprises the following steps: calculating the fault conditions of the distributed power supply and system elements, and calculating the power supply supporting capacity of the distributed power supply during the fault recovery period through time sequence simulation; introducing an island division method based on a forward looking greedy algorithm in the reliability evaluation process; the problem that normally open contact switches are generally ignored in island division research is solved, and the effect of the normally open contact switches with important influence is taken into consideration; and fourthly, continuously searching for an approximate reference value by adopting a string cutting method based on an equal reliability principle in the credible capacity evaluation process to obtain the credible capacity of the distributed power supply with sufficient precision.

The invention solves the practical problem by adopting the following technical scheme:

1. a distributed power supply credible capacity evaluation method based on equal power supply reliability comprises the following steps:

s1, determining the installed capacity of the distributed power supply;

s2, establishing a simulation system model of distributed power supply time sequence output, node time sequence load, system element state and load priority; and according to the simulation system model of the load priority, a 1-NKP model of the islanding problem is established, and the 1-NKP model is solved to obtain a more accurate islanding scheme.

S3, sampling the time sequence states of the distributed power supply and the system element in the evaluation period by adopting a random number sampling method to obtain an operation/fault two-state time sequence model of the system element, and correcting a node injection power curve according to the operation/fault two-state time sequence model of the distributed power supply;

and S4, carrying out islanding on the power distribution network during the fault period, carrying out node voltage and branch load flow verification, and calculating the power supply shortage of each load node and the system according to the obtained islanding scheme.

S5, counting the power shortage information based on the sequential Monte Carlo, and calculating the reliability index of the system;

and S6, carrying out credible capacity search based on the string cutting method to obtain the credible capacity of the distributed power supply.

In step S2, the establishing of the 1-NKP model of the islanding problem according to the simulation system model of the load priority is as follows:

the problem of how the distributed power supply supplies power to the load nodes can be simplified into a knapsack problem. Assuming that there are n relatively independent articles and a backpack with a capacity of C, each article itself has two attributes of weight W and price P, we need to select several of these articles to put into the backpack, so that the sum of the weights of the articles is not greater than the backpack capacity and the final profit is maximized. Similarly, if the vertices in the simple undirected connected graph are considered items, and a vertex is selected for a backpack if at least one of all vertices directly connected to it has been placed in the backpack. This problem is known as 1-NKP. In this problem, the capacity of the distributed power supply can be considered as the backpack capacity C, and the vertex connected to the power supply is considered as the vertex that is first put into the backpack; the power demand of the load node corresponds to the weight W of the item in the backpack problem, and the profit from powering it is noted as P, i.e. the price of the item in the backpack problem. In the case of limited power supply capability C, it is the goal of 1-NKP to determine which load points to supply power to maximize the benefit. And constructing a 1-NKP model of the islanding problem according to the data.

Moreover, the specific method for solving the 1-NKP model in step S2 to obtain a more accurate island division scheme is as follows:

solving a 1-NKP model aiming at the island division problem by adopting a forward-looking greedy algorithm to obtain a more accurate island division scheme, wherein the solving process of the algorithm is as follows:

selecting the vertex where the DG with the maximum power is supplied outwards as the initial point v of 1-NKP₀I.e. Z ═ v₀}; the set Z represents a vertex set drawn into the island, and the outward power supply means that the load value of a vertex where the DG is located is subtracted from the output of the DG at the moment t, namely the power which can be provided for other vertexes connected with the DG by the vertex where the DG is located;

② the sum B of the returns of all the vertexes in the set Z is calculated according to the following equations (6), (7) and (8), respectively_ZSum of total load P_ZAnd DG residual capacity C_R；

B_ZRepresents the sum of the gains of all load points in the current island area:

P_Zrepresenting the total power shortage of all load points in the current island region:

C_Rrepresenting the remaining capacity of the distributed power supply DG:

C_R＝C-P_Z (8)

searching neighborhood of Z and putting neighborhood vertex into set NB¹Performing the following steps;

the neighborhood of a vertex represents a set of vertices adjacent to the vertex, and the neighborhood of a set represents a set of vertices adjacent to the vertex in the set and not belonging to the set; NB¹Neighborhood set, NB, of set Z¹The number of the middle vertexes is marked as X, NB¹(i) Is the set NB¹The ith vertex, i ∈ {1,2, …, X }.

Search for NB¹(i) Is put into the set NB_i ²；

Wherein NB_i ²Representing and vertex NB¹(i) Set of vertices adjacent and not belonging to set Z, i.e. NB¹(i) Can be regarded as the neighborhood of neighborhood points of Z, namely the forward-looking neighborhood of Z. Wherein NB_i ²The number of the middle vertexes is recorded as Y_i，NB_i ²(j) Represents NB¹(i) J is in {1,2, …, X }, j is in {1,2, …, Y }_i}。

Fifthly, calculating the value ratio R according to the following formula (9)_i(j)；

Wherein R is_i(j) Representing a vertex NB¹(i) And its neighborhood point NB_i ²(j) The sum of their gains and the sum of their power consumptions. i ∈ {1,2, …, X }, j ∈ {1,2, …, Y }_i}。

Sixthly, the optimal forward looking value ratio R is obtained according to the following formula (10)²(i)；

Wherein R is²(i) Namely R_i(j) Is called the ith neighborhood point NB of Z¹(i) The "optimal look-ahead value ratio".

Seventhly, calculating the vertex NB according to the following formula (11)¹(i) By itself, value ratio R¹(i)；

B is selected according to the following formula (12)¹Maximum of the intrinsic cost ratio and the optimal lookahead cost ratio among all vertices. If the maximum value is 0, go to step (r); otherwise, recording the vertex corresponding to the maximum value as m. If the value ratios of the different points are the same, the point with the larger profit B is preferentially selected.

Ninthly, adding the vertex m into a set Z, Z ═ Z, m, and judging the DG residual capacity C_RIf it is greater than the power demand P (m), if yes, go back to step two, otherwise execute step R.

Calculating current B in the R_ZAnd P_ZAnd exiting the loop, wherein the elements in the current set Z are the nodes selected by the algorithm to be imported into the island B_ZAnd P_ZThe sum of the profit of the top points in the island and the sum of the required electric quantity are respectively.

The invention has the advantages and beneficial effects that:

1. the method evaluates the credible capacity of the distributed power supply under the criterion of equal power supply reliability, effectively takes the characteristic that the topological structure of the power distribution network can be flexibly changed into consideration, and fully exerts the potential that the distributed power supply can continuously supply power to important loads during the distribution network fault period.

2. When the power supply reliability index of the power distribution network with the distributed power supply is calculated, a 1-NKP island division model is established, and an island division method based on a forward-looking greedy algorithm is introduced, so that the blindness of single-step selection can be overcome, and a better island division scheme can be obtained.

3. In the island dividing process, the important influence of the interconnection switch is taken into account, the power supply priorities of different load nodes are also taken into account, the distributed power supply can be recovered to the important load preferentially during the fault period, the loss caused by power failure is reduced to a greater extent, and the reliability index is more accurate.

Drawings

FIG. 1 is a flow chart of a distributed power source trusted capacity evaluation according to the present invention;

FIG. 2 is a flow chart for making an intelligent islanding scheme of the power distribution network according to the invention;

FIG. 3 is a schematic diagram of the search for confidence capacity based on the chord cutting method according to the present invention.

Detailed description of the preferred embodiments

The embodiments of the invention will be described in further detail below with reference to the accompanying drawings:

a method for evaluating the credible capacity of a distributed power supply of a power distribution network based on equal power supply reliability indexes is shown in figure 1 and comprises the following steps:

step 1, determining installed capacity of a distributed power supply;

the installed capacities of the distributed power supply fan and the photovoltaic are respectively P_wAnd P_pvRepresenting that when calculating the power supply reliability index of the distributed power supply before grid connection, the P is ordered_wAnd P_pvEqual to 0.

In the present embodiment, the installed capacities of the distributed power source fan and the photovoltaic are initialized, which are assumed to be P, respectively_wAnd P_pvLet P_wAnd P_pvEqual to 0.

Step 2, establishing a simulation system model of distributed power supply time sequence output, node time sequence load, system element state and load priority;

the specific steps of the step 2 comprise:

(1) establishing output models of the distributed fans and the photovoltaic, respectively substituting historical data of wind speed and illumination intensity into the output models of the distributed power supply to obtain time sequence output per unit values of the fans and the photovoltaic of each node in an evaluation period, and obtaining time sequence output of the distributed fans and the distributed photovoltaic according to the time sequence output per unit values and the updated installed capacity of the distributed power supply;

the fan and photovoltaic output models are as follows:

a, B, C is a constant in the wind turbine generator output model; SW_tRepresenting the corresponding wind speed at the height of the hub of the wind turbine at the moment t; v_ciFor the cut-in wind speed, V_coFor cut-out wind speed, V_rThe rated wind speed of the fan. T is_c,tRepresents the battery temperature at time t; t is_A,tRepresents the ambient temperature at time t; s_aRepresents the average solar irradiance; k_vRepresents a voltage temperature coefficient; k_iRepresents the temperature coefficient of current; n is a radical of_OTIndicating a standard operating temperature of the battery; FF is the fill factor; i is_sc,tRepresents the short-circuit current at time t; v_oc,tRepresents the open circuit voltage at time t; p_MPPA power representing a maximum power point; i is_MPPA current representing a maximum power point; v_MPPA voltage representing a maximum power point; p_PV,tRepresenting the output power of the photovoltaic module at the moment t; n represents the number of photovoltaic panels.

In the step (2) and the step (1), establishing a fan and photovoltaic output model is an important link for calculating the credible capacity. In a normal working state, the output of the wind turbine generator depends on the meteorological condition, and the output of the fan can be calculated according to the formula (1); the output power of the photovoltaic module depends on the solar irradiance, the site ambient temperature, and the characteristics of the module itself, and can be calculated according to equation (2). Historical data of wind speed and illumination intensity are respectively substituted into the formulas (1) and (2), and the k node fan and the photovoltaic evaluation period T can be obtainedThe time sequence output per unit value of (D) is recorded as

And

according to the updated P_wAnd P_pvRespectively obtaining the time sequence output of the distributed fan and the distributed photovoltaic

And

(2) generating a net load curve of each node by combining load data of each node according to the distributed fan and photovoltaic time sequence output obtained in the step (1) in the step 2;

in this embodiment, according to

And load data of each node, generating a net load curve of each node.

(3) Considering that the power supply recovery values of different load points are different, dividing the loads into three types according to the importance degree, giving different weights, and determining the recovery sequence of the loads according to the weight to obtain a load priority system model.

In this embodiment, when a system fails, power supply is sequentially restored according to the importance of the load. The loads are divided into three types, wherein one type of load is the most important, and the distributed power supply preferentially supplies power to the first type of load after power failure occurs. W (i) is adopted to represent the priority weight of the load point i, and the weight of the first class load, the second class load and the third class load is w respectively₁，w₂And w₃(w₁>w₂>w₃) Thereby measuring the importance of the load point. The higher the load point priority, the greater the corresponding weight.

In this embodiment, the historical data of wind speed and light intensity are respectively substituted into the expressions (1) and (2) to obtain each sectionAnd (3) the time sequence output per unit value of the point fan and the photovoltaic in the evaluation period T, taking the k-th node fan and the photovoltaic as an example, and expressing the time sequence output as

And

per unit value of time sequence output and installed capacity P of distributed power supply_wAnd P_pvObtaining the time sequence output of the distributed fan and the distributed photovoltaic:

and

according to

Generating a net load curve of each node according to the load data of each node; meanwhile, considering that the power supply recovery values of different load points are different, the loads are divided into three types according to the importance degree, different weights are given to the loads, the recovery sequence of the loads is determined according to the weight, and a load priority system model is obtained.

And 3, sampling the time sequence states of the distributed power supply and the system element in the evaluation period by adopting a random number sampling method to obtain an operation/fault two-state time sequence model of the system element, and correcting the node injection power curve according to the operation/fault two-state time sequence model of the distributed power supply.

The specific steps of the step 3 comprise:

(1) assuming that the element is initially in an operational state;

(2) the duration of the current operating state of each element is sampled, taking the operating duration in exponential distribution as an example, assuming λ_iIs the failure rate of the ith element, the working duration of the element is as follows:

wherein, g_iIs [0,1 ]]Random numbers drawn in uniform distribution, resulting in D_iThe value is the working duration of the ith element;

similarly, suppose μ_iIs the repair rate of the ith element, and only the failure rate lambda in the formula (3) is needed_iBecomes the repair rate mu_iAnd the calculated value is the fault repair duration of the ith element.

(3) Repeating the step 3 and the step (2) until the simulation under the research time span is completed;

taking the ith element as an example, let the time span be T, and the fault-free operation time obtained by the jth simulation be T_j,1The component repair duration obtained by the jth simulation is t_j,2(ii) a The simulation total time length t is simulated after the simulation of the u time_zCan be represented by the following formula:

repeating the simulation until the total simulation time t_zStopping the simulation when the research time span T is greater than or equal to the research time span T; generating a timing state of the element over a span of time;

(4) repeating the steps (1) to (3) in the step 3 until the simulation of all the elements in the system is completed;

the outage of the system elements and the distributed power supplies needs to be taken into account simultaneously when evaluating reliability, so that the timing states of all the system elements and the distributed power supplies need to be generated. For the latter, the fan and the photovoltaic power supply have a certain failure rate, and when the device is in a repair state, the output is zero.

(5) And correcting the node injection power curve according to the operation/fault two-state time sequence model of the distributed power supply.

Step 4, according to the operation/fault two-state time sequence model of the system element in the step 3, carrying out power distribution network island division during the fault period; constructing a power distribution network island division optimization model according to the load priority model in the step 2, and solving the model by adopting a forward-looking greedy algorithm to obtain an island division scheme which may contain a ring network; and converting the island division scheme possibly containing the ring network into a tree-shaped island by using a Prim algorithm, checking the voltage and the current of the finally obtained island division scheme, and calculating the power shortage of each load node and the system according to the obtained island division scheme.

The specific steps of the step 4 comprise:

(1) judging whether a fault occurs at the moment t according to the fault condition of the system element obtained by sampling in the step 3, and if no fault occurs, making t equal to t +1, namely jumping to the next moment to continuously judge whether a fault occurs; if the system fails, the next step is entered.

(2) Carrying out intelligent island division on the power distribution network at the time t;

the step 4, the step (2) comprises the following specific steps:

1) according to the fault condition of the system element at the moment t, an undirected graph model G of the power distribution network power failure area at the moment is generated, the area without power failure is not divided into the model G, and a copy G' is saved in the model G.

2) 1-NKP model for establishing island division problem

In this embodiment, the problem of how the distributed power supply supplies power to the load nodes can be simplified into a knapsack problem. Assuming that there are n relatively independent articles and a backpack with a capacity of C, each article itself has two attributes of weight W and price P, we need to select several of these articles to put into the backpack, so that the sum of the weights of the articles is not greater than the backpack capacity and the final profit is maximized. Similarly, if the vertices in the simple undirected connected graph are considered items, and a vertex is selected for a backpack if at least one of all vertices directly connected to it has been placed in the backpack. This problem is known as 1-NKP. In this problem, the capacity of the distributed power supply can be considered as the backpack capacity C, and the vertex connected to the power supply is considered as the vertex that is first put into the backpack; the power demand of the load node corresponds to the weight W of the item in the backpack problem, and the profit from powering it is noted as P, i.e. the price of the item in the backpack problem. In the case of limited power supply capability C, it is the goal of 1-NKP to determine which load points to supply power to maximize the benefit. Accordingly, a 1-NKP model of the islanding problem is constructed:

wherein, x (i) is the state of the load point i, and when the load point i is 1, the load point i is selected into an island, and when the load point i is 0, the load point i is not in the island; w (i) is the priority of load point i; n is the number of load points; p_DGRepresenting the output power limit of the distributed power supply at a certain moment during island operation; p (i) represents the load value at point i; taking the product of P (i) and w (i) as a load i to obtain the income B (i) obtained by the whole power grid after power supply recovery; v represents a vertex set; nb (i) represents the set of neighboring vertices of vertex i in graph G, i.e., the neighborhood of point i; v. of₀Is a DG node; z represents a set of nodes contained in the island region; u shape_bRepresenting the node voltage in the island region; u shape_bminRepresents the lower limit of the node voltage; u shape_bmaxRepresenting the upper node voltage limit. I is_lRepresenting branch current in the island region; i is_lmaxRepresenting the upper branch current limit.

3) Solving the 1-NKP model by adopting a forward-looking greedy algorithm to obtain the island range of the DG power supply with the maximum power supply power, and putting the vertexes contained in the island range into a set S_i。

The specific steps of solving the 1-NKP model by using a forward-looking greedy algorithm in the step 3, the step 2, and the step 3) comprise:

② the sum B of the returns of all the vertexes in the set Z is calculated according to the following equations (6), (7) and (8), respectively_ZThe total amount of the loadAnd P_ZAnd DG residual capacity C_R；

C_Rrepresenting the remaining capacity of the distributed power supply DG:

C_R＝C-P_Z (8)

Search for NB¹(i) Is put into the set NB_i ²；

4) Examination S_iWhether only 1 vertex is included; if yes, representing that the DG can not supply power to the vertex outside the vertex, marking the DG to indicate that the DG is traversed, and then turning to the step 6); otherwise go to step 5).

5) In FIG. G, S_iVertex and S in_iThe edges connected by the other vertices are called "boundary edges", and these boundary edges are denoted as S_iOne end of the inner is deleted and connected to a newly established vertex s_iUsing s_iRepresents S_iAll vertices in (1); load value P of the vertex_siIs S_iThe sum of P of all the vertexes in the system recovers the profit B_siIs S_iSum of B of all vertices in the set, weight w_siIs B_siAnd P_siThe ratio of (A) to (B); s_iIf the vertex contains more DGs, then the DGs are merged into a new DG and connected to s_iThe output power is the sum of the output powers of the DGs. The new graph resulting from this step is still designated as graph G.

6) Checking whether there are any unlabeled DGs in the graph G, if so, turning to step 2); otherwise go to step 7).

7) Each compression point s_iRepresenting an island, point s_iIs reduced to the corresponding set S_iThen, the range of each island is determined from the original graph G', and then all boundary edges are disconnected to form the island which may contain a ring network.

8) And obtaining the edge which needs to be disconnected and converts the island into a radial structure by using a Prim algorithm of a minimum spanning tree.

9) And carrying out voltage constraint and equipment current-carrying capacity constraint verification, and carrying out island adjustment if the constraint is not met.

Based on the steps, the intelligent island division of the power distribution network can be completed, and the process is shown in fig. 2.

(3) And (4) counting the power shortage amount information of the system according to the island division scheme obtained in the step (2) in the step 4, and calculating the reliability index.

The step 4, the step (3) comprises the following specific steps:

1) selectingThe power shortage is selected as a reliability index. At the moment t, according to the island division scheme finally obtained in the step 4 and the step (2), calculating the sum of the power supply shortage of all load nodes in the power distribution network at the moment, and recording the sum as E_1,t. Calculating the electric quantity which can be recovered by the distributed power supply for the power failure load, and recording the electric quantity as E_2,t. At time t, the actual power shortage E of the system_tCan be expressed as:

E_t＝E_1,t-E_2,t (13)

2) judging whether the time T is within the researched time span T, if T is less than or equal to T, returning to the step (1) in the step 4 after T is equal to T + 1; if T > T, the reliability evaluation under the time span T is finished, and the calculation is finished.

According to the operation/fault two-state time sequence model of the system element in the step 3, carrying out power distribution network island division during the fault period; constructing a power distribution network island division optimization model according to the load priority model in the step 2, and solving the model by adopting a forward-looking greedy algorithm to obtain an island division scheme which may contain a ring network; and converting the island division scheme possibly containing the ring network into a tree-shaped island by using a Prim algorithm, checking the voltage and the current of the finally obtained island division scheme, and calculating the power shortage of each load node and the system according to the obtained island division scheme.

Step 5, calculating the reliability index based on the sequential Monte Carlo method

The specific method of the step 5 comprises the following steps:

repeating the steps 3-4 for 1 ten thousand times, counting the reliability index under each simulation, and taking the average value of the power shortage under each island division scheme as the reliability index of the system.

In this embodiment, it is determined whether the monte carlo simulation is repeated 1 ten thousand times. If not, returning to the step 3; and if so, counting the average value of the power shortage amount of the system as the reliability index of the system. The loop is exited.

Step 6, carrying out credible capacity search based on the string cutting method to obtain credible capacity of the distributed power supply

The specific method of the step 6 comprises the following steps:

calculating a reference value R of the reliability index according to the steps 1-5₀And the maximum load of the system is increased to (1+ k) times of the original load. Varying P based on chordal intercept_wAnd P_pvThe step 2 to the step 5 are repeated to calculate the reliability index R of the system until R approaches to R₀And obtaining the credible capacity of the distributed power supply.

In this embodiment, in a certain system, let the capacity of the conventional unit g be C_g，d_tFor the load level of the system at time t, P_re,tAnd G represents the conventional unit set, and R { a, b } represents the reliability of the system under the unit capacity a and the load level b for the output of the distributed power supply re at the time t. Setting the credible capacity of the distributed power supply to be C_cThe trusted capacity evaluation time range is T. Before and after the distributed power supply is connected, the ratio of the difference value of the supplied loads of the system to the wind power installation machine under the same reliability level is the wind power capacity reliability. The calculation criterion for the trusted capacity of the distributed power supply under the definition can be expressed as:

the distributed power supply credibility capacity evaluation method based on the chord cutting method comprises the following specific steps:

(1) calculating a reference value R of the reliability index according to the steps 1-5₀(ii) a And the maximum load of the system is increased to (1+ k) times the original load.

(2) Varying P based on chordal intercept_wAnd P_pvRepeating the steps 2-5 to calculate the reliability index R until R approaches to R₀And obtaining the credible capacity of the distributed power supply.

As shown in fig. 3, the hypothetical curve H is a variation curve of system reliability according to the capacity of the distributed power generator set, the horizontal axis is the installed capacity of the distributed power generator, the vertical axis is the power shortage of the system reliability index, and the larger the numerical value of the reliability index R, the lower the system reliability. The credible capacity of the distributed power supply, namely the corresponding reliability index on the curve H is R₀Point (2) of (c). Assuming the capacity of the distributed power generating unitLimit is 0, upper limit is total capacity p of the unit_maxCalculating the reliability index corresponding to the upper and lower limits of the capacity, finding out the corresponding two points in the graph to connect the two points to obtain a straight line L₁(ii) a Let straight line L₁And R₀The cross axis coordinate corresponding to the intersection point is p₁Calculating the capacity of the distributed power supply unit as p₁The reliability index R of the time system, and then L₂. Repeating the steps to obtain p₂,p₃,p₄… until equation (15) is satisfied, a point p corresponding to the trusted capacity of the distributed power supply is obtained_v。

|R-R₀|＜ε (15)

In the formula, epsilon represents an evaluation convergence criterion of power supply reliability and is a small constant.

The innovation of the invention is that:

1. determining installed capacity of distributed power sources

The installed capacities of the distributed power supply fan and the photovoltaic are respectively P_wAnd P_pvExpressing that P is used for calculating the power supply reliability index of the distributed power supply before grid connection_wAnd P_pvEqual to 0.

2. Model for establishing distributed power supply, power grid element and load in system

And (3) constructing a fan and photovoltaic output model, and obtaining a time sequence output curve of the distributed power supply according to factors such as wind speed and illumination intensity.

3. Generating system time sequence operation state

Sampling the time sequence states of the distributed power supply and the system elements in the evaluation period T based on a random number sampling method to obtain an operation/fault two-state time sequence model of the system elements, and correcting the node injection power curve according to the operation/fault two-state time sequence model of the distributed power supply.

4. Fault consequence analysis based on intelligent island division

5. Calculating reliability index based on sequential Monte Carlo method

Repeating the step 3 and the step 4 for 1 ten thousand times, counting the reliability index under each simulation, and taking the average value of the power shortage under each island division scheme as the reliability index of the system.

6. Credible capacity search based on string-cutting method

Marking the reliability index of the distributed power supply when the installed capacity is zero as a reference value R₀. Increasing the maximum load of the system to (1+ k) times of the original load, and changing P based on the chord cutting method_wAnd P_pvThe step 2 to the step 5 are repeated to calculate the reliability index R of the system until R approaches to R₀And obtaining the credible capacity of the distributed power supply.

The working principle of the invention is as follows:

the invention provides a credible capacity evaluation method for a distributed power supply, which effectively considers the characteristic that the topological structure of a power distribution network can be flexibly changed, can fully exert the potential of the distributed power supply for continuously supplying power to important loads during the distribution network fault period, evaluates the credible capacity of the distributed power supply under the criterion of equal power supply reliability, and improves the accuracy of credible capacity evaluation. The reliability calculation of the system is a core link of the credible capacity evaluation of the distributed power supply, and when the power supply reliability index of the power distribution network containing the distributed power supply is calculated, the power supply potential of the distributed power supply during the fault recovery period is fully excavated through dynamic island division. When the island division optimization model is established, the important influence of a tie switch is considered, the power supply priorities of different load nodes are also considered, and a prospective greedy algorithm is introduced to rapidly solve the island division optimization model. And finally, taking the installed capacity of the distributed power supply as an optimization object, and searching the installed capacity which enables the reliability index of the system to be equal before and after the distributed power supply is connected based on a string cutting method to obtain a reliable capacity result with enough precision.

It should be emphasized that the examples described herein are illustrative and not restrictive, and thus the present invention includes, but is not limited to, those examples described in this detailed description, as well as other embodiments that can be derived from the teachings of the present invention by those skilled in the art and that are within the scope of the present invention.

Claims

1. A distributed power supply credible capacity evaluation method based on equal power supply reliability is characterized by comprising the following steps: the method comprises the following steps:

s1, determining the installed capacity of the distributed power supply;

s2, establishing a simulation system model of distributed power supply time sequence output, node time sequence load, system element state and load priority; establishing a 1-NKP model of the islanding problem according to the simulation system model of the load priority, and solving the 1-NKP model to obtain a more accurate islanding scheme;

s4, carrying out islanding on the power distribution network during the fault period, carrying out node voltage and branch load flow verification, and calculating the power supply shortage of each load node and the system according to the obtained islanding scheme;

2. The distributed power supply credible capacity evaluation method based on equal power supply reliability as claimed in claim 1, characterized in that: the step S2 of establishing the 1-NKP model of the islanding problem according to the simulation system model of the load priority is:

the problem of how the distributed power supply supplies power to the load nodes can be simplified into a knapsack problem; supposing that n relatively independent articles and a backpack with the capacity of C are provided, each article has two attributes of weight W and price P, and a plurality of articles need to be selected from the articles and put into the backpack, so that the sum of the weights of the articles is not more than the capacity of the backpack and the final benefit is the maximum; similarly, if the vertices in the simple undirected connected graph are considered items, and a vertex is selected for a backpack if at least one of all vertices directly connected to it has been placed in the backpack; such problems are known as 1-NKP; in this problem, the capacity of the distributed power supply can be considered as the backpack capacity C, and the vertex connected to the power supply is considered as the vertex that is first put into the backpack; the power demand of the load node is equivalent to the weight W of the goods in the knapsack problem, and the income brought by supplying power to the load node is marked as P, namely the price of the goods in the knapsack problem; in the case of limited power supply capacity C, how to determine which load points to supply power so that the maximum benefit is the target of 1-NKP; and constructing a 1-NKP model of the islanding problem according to the data.

3. The distributed power supply credible capacity evaluation method based on equal power supply reliability as claimed in claim 1, characterized in that: the specific method for solving the 1-NKP model in step S2 to obtain a more accurate island division scheme is as follows:

② the sum B of the returns of all the vertexes in the set Z is calculated according to the following equations (6), (7) and (8), respectively_ZTotal amount of loadSum P_ZAnd DG residual capacity C_R；

C_Rrepresenting the remaining capacity of the distributed power supply DG:

C_R＝C-P_Z (8)

the neighborhood of a vertex represents a set of vertices adjacent to the vertex, and the neighborhood of a set represents a set of vertices adjacent to the vertex in the set and not belonging to the set; NB¹Neighborhood set, NB, of set Z¹The number of the middle vertexes is marked as X, NB¹(i) Is the set NB¹The ith vertex, i ∈ {1,2, …, X };

search for NB¹(i) Is put into the set NB_i ²；

Wherein NB_i ²Representing and vertex NB¹(i) Set of vertices adjacent and not belonging to set Z, i.e. NB¹(i) The neighborhood of (2) can be regarded as a neighborhood of a neighborhood point of Z, namely a forward-looking neighborhood of Z; wherein NB_i ²The number of the middle vertexes is recorded as Y_i，NB_i ²(j) Represents NB¹(i) J is in {1,2, …, X }, j is in {1,2, …, Y }_i}；

Wherein R is_i(j) Representing a vertex NB¹(i) And its neighborhood point NB_i ²(j) The ratio of the sum of the gains of (a) to the sum of their power consumptions; i ∈ {1,2, …, X }, j ∈ {1,2, …, Y }_i}；

Wherein R is²(i) Namely R_i(j) Is called the ith neighborhood point NB of Z¹(i) The "optimal look-ahead value ratio";

B is selected according to the following formula (12)¹Maximum value of the value ratio of the vertex and the optimal forward-looking value ratio; if the maximum value is 0, go to step (r); otherwise, recording the vertex corresponding to the maximum value as m; if the value ratios of different points are the same, preferentially selecting the point with larger income B;

ninthly, adding the vertex m into a set Z, Z ═ Z, m, and judging the DG residual capacity C_RIf the power is larger than the required power P (m) of point m, returning to step II, otherwise executing step R;